Full Review,cyclic peptides with non-canonical amino acids

The field of biomolecular design is undergoing a significant transformation, driven by the increasing exploration and application of non-canonical cyclic peptides. These complex molecules, characterized by their unique cyclic structure and the incorporation of non-canonical amino acids (ncAAs), are expanding the functional chemical space available for a wide range of applications, from drug discovery to advanced materials. Understanding the intricacies of non-canonical cyclic peptides is crucial for researchers and developers seeking to harness their full potential.

Cyclic peptides themselves offer inherent advantages, including enhanced structural stability, resistance to proteolytic degradation, and potentially improved oral bioavailability, making them promising drug candidates. The addition of non-canonical amino acids further amplifies these benefits by introducing novel chemical functionalities and structural constraints. This integration allows for the fine-tuning of properties such as binding affinity, specificity, and pharmacokinetic profiles. The study of cyclic peptide structure prediction and design using AlphaFold2, and more recently AlphaFold3, highlights the growing need for accurate computational tools to model these intricate structures.

Recent advancements in artificial intelligence have been pivotal in this domain. Novel deep learning models, such as NCPepFold, have been designed specifically for cyclic peptides with noncanonical amino acids and can operate without length restrictions. These models are revolutionizing peptide structure prediction, offering unprecedented accuracy in modeling noncanonical cyclic peptides. Similarly, AlphaFold3 is being actively explored for its capabilities in cyclic peptide structure prediction, with ongoing research focusing on its application to noncanonical cyclic peptides. While AlphaFold3 has demonstrated remarkable success in protein structure prediction, its specific performance with non-canonical cyclic peptides is an active area of investigation. Emerging tools like NCFlow are also showing promise, with studies indicating that NCFlow outperforms AlphaFold3-based methods in the structure prediction of unseen non-canonical amino acids, suggesting a dynamic and evolving landscape of predictive technologies.

The incorporation of non-canonical amino acids is not a new concept but has seen a surge in interest due to its ability to engineer peptides and proteins with novel properties. These ncAAs are increasingly used to expand the functional chemical space available to biomolecular design. Researchers are developing sophisticated methods for their parameterizing non-canonical amino acids for cyclic peptide simulations, enabling higher-quality ring seeding for computational modeling. This is essential for accurate predictions and for understanding the conformational landscapes of these molecules. Moreover, the total synthesis and biological evaluation of natural cyclic peptides continues to be a vital area of research, providing inspiration and benchmarks for synthetic efforts.

Datasets are also evolving to support this research. Initial cyclic peptide structures are being curated from existing databases like CREMP, while noncanonical residue structures are sourced from repositories such as the RCSB. This data is foundational for training and validating predictive models. The development of platforms like PepLand, a large-scale pre-trained peptide representation model, further aids in analyzing cyclic peptides, which are polypeptides containing non- canonical amino acids, due to their promising therapeutic potential.

The challenges and prospects of using non-canonical amino acids in peptide drug discovery are significant. The inspiration for applying ncAAs to macrocyclic peptide drug discovery often stems from the diverse structures found in natural peptide products. Researchers are exploring various strategies, including genetically programmed cell-based synthesis to create non-natural macrocyclic peptides, each incorporating multiple non-canonical amino acids. These advances are paving the way for the development of cyclic peptides with tailored functionalities, such as non-competitive cyclic peptides for targeting enzyme active sites.

The pursuit of accurate structure prediction for non-canonical cyclic peptides remains a key focus. Models like CyclicBoltz1, derived from AlphaFold 3 variants, are designed for fast and accurate prediction of cyclic peptides and complexes containing non-canonical amino acids. The evaluation of confidence metrics for non-canonical cyclic peptides is also crucial to ensure the reliability of predicted structures.

In summary, the field of non-canonical cyclic peptides is a rapidly advancing area within biomolecular science. The synergistic combination of innovative computational tools, a deeper understanding of non-canonical amino acids, and sophisticated synthetic methodologies is unlocking the immense potential of these molecules. Whether for therapeutic applications or for the broader expansion of chemical diversity, cyclic peptides with their non-canonical components represent a frontier of discovery.